The frameworks aren't theory. Most of them are running software.

The SFW Filter became a deck analyzer

In 2008 I took a red pen to a forty-five-slide deck and asked every slide what it was for. Seventeen years later I built the machine that does it.

It reads a client-facing deck, kills filler, rewrites weak headlines, and turns data into decisions. If a slide can't answer so f***ing what?, it doesn't belong.

It'll be free, in the library, when the book lands.

The Ceiling Question and the Bleeding Taxonomy became a media planner

What would have to be true to make this possible? is a question in the book. In a media plan it's arithmetic. The system takes performance data and models what changes at different spend levels, the Ceiling Question expressed in dollars instead of language.

The Bleeding Taxonomy sits underneath it. Media waste is the canonical slow bleed: a few hundred dollars a day, invisible until somebody totals it.

The Telephone Game and the 87% Rule became the Brief Builder

Chapter 7 diagnoses it: strategy degrades with every handoff. Chapter 8 fixes it, with seven questions a brief has to answer before it's allowed to leave the building.

The Brief Builder is that chapter as a working tool. It interviews you, catches the vague answers that would doom a brief, and hands back a document ready for review.

Also free, also in the library.

Three that don't have tools, and shouldn't

The Storm Principle, the Momentum Principle and the Smarter Fella Test have no software behind them.

They're about what a person does when the situation is uncomfortable: walk into the weather, ship before it's perfect, imagine somebody better. You can't automate nerve.

And one that's inside all of them

The Columbo Strategy isn't a tool. It's a way of listening: the person asking questions controls the room.

But it's in everything above. The brief builder interviews you and refuses to accept a vague answer. The prospect dossier is a list of questions somebody should have asked before the pitch. The deck analyzer asks every slide what it's for and won't take because it's important for an answer.

It didn't become software. It's the reason the software behaves the way it does.

The Content Reactor

Most organizations don't have a content problem. They have a throughput problem. The ideas are in voice notes, Slack threads that scroll into oblivion, half-written docs, and the founder's head, where they compete with everything else for shelf space. Nothing is catching them.

Voice memo in. Finished output out, in the client's voice rather than AI-generic. One idea becomes a post, a newsletter, a carousel and a thread.

A week of content in about forty-five minutes, instead of the hours of writing it replaces.

The test it had to pass: does this produce content that answers so f***ing what?, or generic slop that fills a calendar?

Prospect intelligence

Most agencies lose deals before the proposal is written. Research gets rushed, rebuilt from scratch every time, or quietly skipped, and the team walks in with a LinkedIn scroll and a prayer.

Structured dossiers instead: who the decision-makers are and what they care about, what's pressuring them competitively, what just changed in earnings and hiring, and the angles that follow.

Minutes, instead of the days of desk research it replaces.

The test: does this give you a point of view, or just a pile of data?

Client reporting, all of it

Every reporting system I run is built on the same method. Not a dashboard, not a fifty-page PDF. A brief that says what changed, why it matters, and what to do about it, delivered where people already are.

The test: does this tell someone what to DO, or just what happened?

SFWRecon

The prospect intelligence work turned into a product. A prospect-intelligence platform I built and am still shaping. The same question, pointed at a market instead of a client.

A decision database, not a notes file

Everybody is building "AI memory" and most of them mean a chatbot with a notes file. I built something that records what was decided, why, and what supersedes it.

It's the reason I can tell you what I chose eight months ago and what I was wrong about.

The test: does this remember what was decided and why, or just what was said?

A morning brief that reads the comment, not the checkbox

The operations brief that starts my day reads what my team actually wrote in the weekly client note. Not whether the task got ticked.

A checkbox tells you somebody did a thing. The comment tells you what they found. Almost every status system ever built reports the first one, because it's easier and it looks like information.

The test: does this tell me what's happening, or that somebody said something was happening?

A grant system for a nonprofit that runs on federal funding

One person holds the institutional knowledge for the whole grant cycle. When a renewal comes up she opens ten prior applications side by side and copy-pastes her way to a first draft she describes, accurately, as terrible. Then she rewrites it. The organization's funding depends on a process that lives mostly in one person's head.

The obvious build is retrieval: point it at the archive, let it assemble drafts. That's what the engagement was scoped as.

Discovery turned up something better. After every decision the funder sends back a document scoring the application section by section, with a written account of what the reviewers thought was strong and what wasn't. Most organizations file it and move on. It's the answer key, and nobody was writing against it.

The test it has to pass: does this produce an application that satisfies the instructions, or one that would score well?

Those are different systems. The second one is worth building.

In design. First live use is a January federal deadline the client set, not one we did.

An attention system for three hundred ad accounts

A platform business runs paid search across around three hundred client accounts. Nobody can look at three hundred accounts in a week. Nobody can look at thirty.

The instinct is to build reporting: pull the data, generate summaries, deliver on a schedule. That produces three hundred reports a week and no more attention than there was before. It moves the bottleneck from looking at accounts to looking at reports about accounts.

The real problem isn't reporting. It's triage. Of three hundred accounts, which four need a human this week, and why?

The test: does this tell someone which accounts to look at, or produce three hundred reports nobody reads?

In design. Deferred once already, deliberately, in favour of work that had to come first.